COMP4020: Agentic Coding Studio
Twelve weeks of shipping small full-stack prototypes with an AI coding agent doing most of the typing — the actual skill being assessed is direction, grounding and correction, not lines of code. Static sites for the first half, databases and deploys for the second. This entry is the one honest bit of self-reference in the catalog.
Credit units
6 units
Instructor
Dr. Morgan Reyes
Prerequisites
COMP2100, or equivalent experience building and shipping a non-trivial program.
Learning objectives
By the end of this course, students will be able to:
- Direct an AI coding agent toward a working prototype from an ambiguous brief
- Ground an agent's output against real constraints instead of accepting it uncritically
- Recover cleanly when an agent's approach goes wrong
- Present a week's build and defend the decisions behind it
Assessment
- Weekly crit submissions60%Ten weekly prototypes, each briefly presented and defended in a crit session; graded on direction and correction, not lines of code.
- Process journal15%A running record of prompts, decisions and corrections kept across the semester.
- Final showcase25%A polished version of one semester project, deployed and demoed live.
No written exam — assessment is entirely built artefacts, live crits and a final demo.
Timetable
Lecture: Mon 2pm–3:30pm, Hanna Neumann G51
2 tutorial option(s) — pick one on the timetable page after enrolling.
- Crit slot A: Mon 3:30pm–5pm, Hanna Neumann G51
- Crit slot B: Thu 1pm–2:30pm, Hanna Neumann G52
4/4 courses added.